Hey, thanks for your response. I’ll give my thoughts on your answers, almost all of which I agree with. And I’ll give a bit of info from my own research into phantom entries, etc.
My understanding is that editing existing items results in the database containing not only the revised item but also the corresponding item before the edit was made.
This is my understanding, too. I imagine this bug arises because it is computationally expensive to check every level to ensure the phantom entry is no longer used in any other level after being removed from a particular level. This could’ve been fixed a long time ago by implementing each database entry as a one-to-many field (connecting words in the database to words in the levels, which can repeat) and then modifying these lists as the word is added or removed in levels. Because of the existence of this bug, I assume it wasn’t implemented this way, or it would likely be trivial to fix. Fixing this now on Memrise’s side would likely be very computationally expensive, effectively requiring a refactor of all community courses. Or a manually-triggered one-by-one refactor of courses, triggered by creators/contributors.
The other possibility is that they keep it in the database in case you accidentally deleted something that you didn’t mean to. This would be a rather unusual design decision, though. (And even if this is the case, this could also be fixed with the one-to-many idea, since you could only pull database entries with a usage count greater than 0 when selecting for multiple choice). Overall, there’s little excuse for this bug unless they were trying to save server space by not turning each word into a list…
This is a longstanding Memrise bug that course creators (and contributors) have no control over.
There is a little bit of control, in that you can (carefully) delete them from the database.
While researching now, the least time-consuming method I’ve found to deal with them is to follow a procedure similar to the one outlined here. And the second post in that thread includes a means of doing so without resetting everyone’s progress.
and there have been times when the script stopped working due to Memrise updates.
That’s very, very true. Updates have very regularly broken userscripts over the years. And so I don’t know the state of it either. I was mostly linking it as proof that contributors could have control over the database and the bug, rather than just course creators. If contributors were entirely locked out of editing the database outside of the level editor this bug would be much more severe.
- For duplicates with minor differences, you could painstakingly edit these (or delete duplicates) using the course edit mode. But then new, corresponding “phantom entries” will start appearing anyway during learning sessions.
I don’t think this is necessarily the case as long as you then delete the old version of the word from the database afterwards. This certainly adds an extra step, though. But I believe it should be possible to save that extra step until the end and handle it via one of the above procedures.
You could also split items with more than one Spanish word. But this will likely confuse existing users.
I actually don’t mind having multiple words as a single flashcard. I just want them to be standardized and indicated in some format. Perhaps for indication, a number in brackets next to the prompt word (or, ideally, a different field entirely but this would require Ortegam since contributors cannot edit the fields of the database).
And as for standardization: have the synonyms in alphabetical order; with a comma and single space between synonyms; and with equivalent means of indicating gender, plurality, and/or reflexivity, etc. between synonyms, except where differences are necessary (a translation that is always plural, for example).
I think this change in particular would make the course very useful for studying synonyms, especially for those that study on computer with typed answers. (This is how I always study. I don’t use the app. I figure this is probably rather rare, though).
I would suggest changing all nouns to include the definite article (e.g. “el” before the noun instead of " (m.) "after the noun.
I agree with this suggested change, and I agree with using Vim or similar to do so. It would still need to be done carefully to ensure nothing gets wrongly replaced, though (the classic Dawizard problem). And because this can differ so heavily from word-to-word (some have (m), (m ), (m pl), (m/pl), typos such as (m0, and feminine variants etc). It would likely need a regular expression of some kind. Definitely something that should be done with multiple backups and a careful manual double-checking.
The Ortegam course is fairly old, and is almost certainly based on a 10+ year old Edexcel specification. There is a new draft specification being released by Edexcel on 13 July 2023 that might be of interest to you.
This is very helpful, thank you. I’d either update the course to better fit this format (adding additional words/levels if necessary, while still keeping as much of the existing content as possible, even if rearranged) or change the name to something along the lines of “Spanish GCSE Supplement” to make it obvious that it’s not up-to-date and shouldn’t be used as the sole means of studying.
Let me know if more commentary from me would be helpful.
Your answers have been helpful, thank you. I hope my responses here are also of interest to you as well. And let me know if anything I’ve said here sounds wrong or incomplete or if you have additional info you think would be of use.